path: root/drivers/gpu/drm/vc4/vc4_drv.h

/* SPDX-License-Identifier: GPL-2.0-only */
/*
 * Copyright (C) 2015 Broadcom
 */
#ifndef _VC4_DRV_H_
#define _VC4_DRV_H_

#include <linux/debugfs.h>
#include <linux/delay.h>
#include <linux/of.h>
#include <linux/refcount.h>
#include <linux/uaccess.h>

#include <drm/drm_atomic.h>
#include <drm/drm_debugfs.h>
#include <drm/drm_device.h>
#include <drm/drm_encoder.h>
#include <drm/drm_fourcc.h>
#include <drm/drm_gem_dma_helper.h>
#include <drm/drm_managed.h>
#include <drm/drm_mm.h>
#include <drm/drm_modeset_lock.h>

#include <kunit/test-bug.h>

#include "uapi/drm/vc4_drm.h"

struct drm_device;
struct drm_gem_object;

extern const struct drm_driver vc4_drm_driver;
extern const struct drm_driver vc5_drm_driver;

/* Don't forget to update vc4_bo.c: bo_type_names[] when adding to
 * this.
 */
enum vc4_kernel_bo_type {
	/* Any kernel allocation (gem_create_object hook) before it
	 * gets another type set.
	 */
	VC4_BO_TYPE_KERNEL,
	VC4_BO_TYPE_V3D,
	VC4_BO_TYPE_V3D_SHADER,
	VC4_BO_TYPE_DUMB,
	VC4_BO_TYPE_BIN,
	VC4_BO_TYPE_RCL,
	VC4_BO_TYPE_BCL,
	VC4_BO_TYPE_KERNEL_CACHE,
	VC4_BO_TYPE_COUNT
};

/* Performance monitor object. The perform lifetime is controlled by userspace
 * using perfmon related ioctls. A perfmon can be attached to a submit_cl
 * request, and when this is the case, HW perf counters will be activated just
 * before the submit_cl is submitted to the GPU and disabled when the job is
 * done. This way, only events related to a specific job will be counted.
 */
struct vc4_perfmon {
	struct vc4_dev *dev;

	/* Tracks the number of users of the perfmon, when this counter reaches
	 * zero the perfmon is destroyed.
	 */
	refcount_t refcnt;

	/* Number of counters activated in this perfmon instance
	 * (should be less than DRM_VC4_MAX_PERF_COUNTERS).
	 */
	u8 ncounters;

	/* Events counted by the HW perf counters. */
	u8 events[DRM_VC4_MAX_PERF_COUNTERS];

	/* Storage for counter values. Counters are incremented by the HW
	 * perf counter values every time the perfmon is attached to a GPU job.
	 * This way, perfmon users don't have to retrieve the results after
	 * each job if they want to track events covering several submissions.
	 * Note that counter values can't be reset, but you can fake a reset by
	 * destroying the perfmon and creating a new one.
	 */
	u64 counters[] __counted_by(ncounters);
};

enum vc4_gen {
	VC4_GEN_4,
	VC4_GEN_5,
	VC4_GEN_6_C,
	VC4_GEN_6_D,
};

struct vc4_dev {
	struct drm_device base;
	struct device *dev;

	enum vc4_gen gen;

	unsigned int irq;

	struct vc4_hvs *hvs;
	struct vc4_v3d *v3d;

	struct vc4_hang_state *hang_state;

	/* The kernel-space BO cache.  Tracks buffers that have been
	 * unreferenced by all other users (refcounts of 0!) but not
	 * yet freed, so we can do cheap allocations.
	 */
	struct vc4_bo_cache {
		/* Array of list heads for entries in the BO cache,
		 * based on number of pages, so we can do O(1) lookups
		 * in the cache when allocating.
		 */
		struct list_head *size_list;
		uint32_t size_list_size;

		/* List of all BOs in the cache, ordered by age, so we
		 * can do O(1) lookups when trying to free old
		 * buffers.
		 */
		struct list_head time_list;
		struct work_struct time_work;
		struct timer_list time_timer;
	} bo_cache;

	u32 num_labels;
	struct vc4_label {
		const char *name;
		u32 num_allocated;
		u32 size_allocated;
	} *bo_labels;

	/* Protects bo_cache and bo_labels. */
	struct mutex bo_lock;

	/* Purgeable BO pool. All BOs in this pool can have their memory
	 * reclaimed if the driver is unable to allocate new BOs. We also
	 * keep stats related to the purge mechanism here.
	 */
	struct {
		struct list_head list;
		unsigned int num;
		size_t size;
		unsigned int purged_num;
		size_t purged_size;
		struct mutex lock;
	} purgeable;

	uint64_t dma_fence_context;

	/* Sequence number for the last job queued in bin_job_list.
	 * Starts at 0 (no jobs emitted).
	 */
	uint64_t emit_seqno;

	/* Sequence number for the last completed job on the GPU.
	 * Starts at 0 (no jobs completed).
	 */
	uint64_t finished_seqno;

	/* List of all struct vc4_exec_info for jobs to be executed in
	 * the binner.  The first job in the list is the one currently
	 * programmed into ct0ca for execution.
	 */
	struct list_head bin_job_list;

	/* List of all struct vc4_exec_info for jobs that have
	 * completed binning and are ready for rendering.  The first
	 * job in the list is the one currently programmed into ct1ca
	 * for execution.
	 */
	struct list_head render_job_list;

	/* List of the finished vc4_exec_infos waiting to be freed by
	 * job_done_work.
	 */
	struct list_head job_done_list;
	/* Spinlock used to synchronize the job_list and seqno
	 * accesses between the IRQ handler and GEM ioctls.
	 */
	spinlock_t job_lock;
	wait_queue_head_t job_wait_queue;
	struct work_struct job_done_work;

	/* Used to track the active perfmon if any. Access to this field is
	 * protected by job_lock.
	 */
	struct vc4_perfmon *active_perfmon;

	/* The memory used for storing binner tile alloc, tile state,
	 * and overflow memory allocations.  This is freed when V3D
	 * powers down.
	 */
	struct vc4_bo *bin_bo;

	/* Size of blocks allocated within bin_bo. */
	uint32_t bin_alloc_size;

	/* Bitmask of the bin_alloc_size chunks in bin_bo that are
	 * used.
	 */
	uint32_t bin_alloc_used;

	/* Bitmask of the current bin_alloc used for overflow memory. */
	uint32_t bin_alloc_overflow;

	/* Incremented when an underrun error happened after an atomic commit.
	 * This is particularly useful to detect when a